Finding over and under-represented interactions using HOMER

Question

[pixie@bioinfo]

I am analyzing Hi-C data with HOMER. My experimental set up is time-point wise analysis of stress (0,1 and 6 hrs). For a differential analysis, say 6 hrs vs 0 hr, how can I find interactions which are over-represented, under_represented and specific to only one condition ? I had initially just used the FDR values to fish out significant interactions. There is a confusion in my lab regarding this issue and we are stuck at a critical point. Kindly help.

I have the following important columns in the interaction table for 6 hrs vs 0 hr:

HiCseq_result_InteractionID

HiCseq_result_Total.Read.1

HiCseq_result_Total.Read.2

HiCseq_result_Z.score

HiCseq_result_LogP

HiCseq_result_FDR.Benjamini

HiCseq_result_Bg.Interaction.Reads

HiCseq_result_Bg..2pected.Reads

HiCseq_result_Bg.Z.score

HiCseq_result_Bg.LogP

HiCseq_result_Bg.Total.Reads.Peak1

HiCseq_result_Bg.Total.Reads.Peak2

HiCseq_result_LogP.vs..Bg

HiCseq_result_Z.score.Difference.vs..Bg

[ATpoint]

Not familiar with HOMER's ability towards HiC but I suggest you use a proper statistical framework for this such as diffHiC. Like all tools from Aaron Lun the documentation is extensive and high quality.